Outlie Itroductio Backgroud o Distributed DBMS Architecture Datalogical Architecture Implemetatio Alteratives Compoet Architecture o Distributed DBMS Architecture o Distributed Desig o Sematic Data Cotrol o Distributed Query Processig o Distributed Trasactio Maagemet o Parallel Systems o Distributed Object DBMS o Iteroperability o Curret Issues Page 4. 1 Architecture Defies the structure of the system compoets idetified fuctios of each compoet defied iterrelatioships ad iteractios betwee compoets defied Page 4. 2 Page 1
ANSI/SPARC Architecture Users Exteral Exteral view Exteral view Exteral view Coceptual Coceptual view Iteral Iteral view Page 4. 3 Stadardizatio Referece Model A coceptual framework whose purpose is to divide stadardizatio work ito maageable pieces ad to show at a geeral level how these pieces are related to oe aother. Approaches Compoet-based u Compoets of the system are defied together with the iterrelatioships betwee compoets. u Good for desig ad implemetatio of the system. Fuctio-based u Classes of users are idetified together with the fuctioality that the system will provide for each class. u The objectives of the system are clearly idetified. But how do you achieve these objectives? Data-based u Idetify the differet types of describig data ad specify the fuctioal uits that will realize ad/or use data accordig to these views. Page 4. 4 Page 2
Coceptual Defiitio RELATION EMP [ KEY = {ENO} ATTRIBUTES = { ENO : CHARACTER(9) ENAME : CHARACTER(15) TITLE : CHARACTER(10) } ] RELATION PAY [ KEY = {TITLE} ATTRIBUTES = { TITLE : CHARACTER(10) SAL : NUMERIC(6) } ] Page 4. 5 Coceptual Defiitio RELATION PROJ [ KEY = {PNO} ATTRIBUTES = { PNO : CHARACTER(7) PNAME : CHARACTER(20) BUDGET : NUMERIC(7) LOC : CHARACTER(15) } ] RELATION ASG [ KEY = {ENO,PNO} ATTRIBUTES = { ] } ENO PNO RESP DUR : CHARACTER(9) : CHARACTER(7) : CHARACTER(10) : NUMERIC(3) Page 4. 6 Page 3
Iteral Defiitio RELATION EMP [ KEY = {ENO} ATTRIBUTES = { ] } ENO ENAME TITLE : CHARACTER(9) : CHARACTER(15) : CHARACTER(10) INTERNAL_REL E [ INDEX ON E# CALL EMINX FIELD = { ] } E# : BYTE(9) ENAME : BYTE(15) TIT : BYTE(10) Page 4. 7 Exteral View Defiitio Ð Example 1 Create a BUDGET view from the PROJ relatio CREATE VIEW BUDGET(PNAME, BUD) AS SELECT PNAME, BUDGET FROM PROJ Page 4. 8 Page 4
Exteral View Defiitio Ð Example 2 Create a Payroll view from relatios EMP ad TITLE_SALARY CREATE VIEW PAYROLL (EMP_NO, EMP_NAME, SAL) AS SELECT EMP.ENO,EMP.ENAME,PAY.SAL FROM EMP, PAY WHERE EMP.TITLE = PAY.TITLE Page 4. 9 DBMS Implemetatio Alteratives Page 4. 10 Page 5
Dimesios of the Problem Distributio Whether the compoets of the system are located o the same machie or ot Heterogeeity Various levels (hardware, commuicatios, operatig system) DBMS importat oe u data model, query laguage,trasactio maagemet algorithms Autoomy Not well uderstood ad most troublesome Various versios [Veijalaie ad Pepescu-Zeleti, 1988] u Desig autoomy: Ability of a compoet DBMS to decide o issues related to its ow desig. u Commuicatio autoomy: Ability of a compoet DBMS to decide whether ad how to commuicate with other DBMSs. u Executio autoomy: Ability of a compoet DBMS to execute local operatios i ay maer it wats to. Page 4. 11 Datalogical Distributed DBMS Architecture... ES 1 ES 2 ES GCS... LCS 1 LCS 2 LCS... LIS 1 LIS 2 LIS Page 4. 12 Page 6
Datalogical Multi-DBMS Architecture GES 1 GES 2... GES É É LES 11 LES 1 GCS LES 1 LES m LCS 1 LCS 2 É LCS LIS 1 LIS 2 É LIS Page 4. 13 Timesharig Access to a Cetral No data storage Host ruig all software Termials or PC termial emulators Batch Respose requests Applicatio Software DBMS Services Network Page 4. 14 Page 7
Multiple Cliets/Sigle Server Applicatios Cliet Services Applicatios Cliet Services Applicatios Cliet Services High-level requests Filtered data oly LAN DBMS Services Page 4. 15 Task Distributio Applicatio QL Iterface Programmatic Iterface Maager SQL query Maager Query Optimizer result table Lock Maager Storage Maager Page & Cache Maager Page 4. 16 Page 8
Advatages of Cliet-Server Architectures More efficiet divisio of labor Horizotal ad vertical scalig of resources Better price/performace o cliet machies Ability to use familiar tools o cliet machies Cliet access to remote data (via stadards) Full DBMS fuctioality provided to cliet workstatios Overall better system price/performace Page 4. 17 Problems With Multiple- Cliet/Sigle Server Server forms bottleeck Server forms sigle poit of failure scalig difficult Page 4. 18 Page 9
Multiple Cliets/Multiple Servers directory cachig query decompositio commit protocols Applicatios Cliet Services LAN DBMS Services DBMS Services Page 4. 19 Server-to-Server SQL iterface programmatic iterface other applicatio support eviromets Applicatios Cliet Services LAN DBMS Services DBMS Services Page 4. 20 Page 10
Peer-to-Peer Compoet Architecture USER PROCESSOR DATA PROCESSOR User requests USER Exteral User Iterface Hadler Sematic Data Cotroller Global Coceptual Global Query Optimizer GD/D Global Executio Moitor Local Coceptual Local Query Processor System Log Local Recovery Maager Local Iteral Rutime Support Processor System resposes Page 4. 21 Compoets of a Multi-DBMS Resposes USER Global Requests GTP GS GUI GRM GQP GQO Local Requests Compoet Iterface Processor (CIP) Compoet Iterface Processor (CIP) Local Requests D B M S User Iterface Query Processor Query Optimizer Rutime Sup. Processor Trasactio Maager Scheduler Recovery Maager É D B M S Trasactio Maager Scheduler Recovery Maager Rutime Sup. Processor User Iterface Query Processor Query Optimizer Page 4. 22 Page 11
Directory Issues Type Global & cetral & o-replicated Local & cetral & o-replicated (?) Local & distributed & o-replicated Local & cetral & replicated (?) Global & distributed & o-replicated (?) Global & cetral & replicated (?) Replicatio Global & distributed & replicated Local & distributed & replicated Locatio Page 4. 23 Page 12